A hybrid modeling method of metal hydride tank and dynamic characteristic analysis

To design and optimize the thermal management of metal hydride (MH) tank, high-fidelity model (HFM) composed of numerous partial differential equations, is often adopted to capture the spatial evolution of key parameters accurately. However, computational difficulties raise in dealing with such nonlinear interconnections for system-level simulation. Therefore, the model reduction of MH system has drawn much attention aiming at the trade-off between accuracy and computational cost. In this paper, a hybrid modeling method which integrates advantages of the state-space model (SSM) and HFM, is proposed and applied for the model reduction of MH tank. In particular, four data-driven matrices are introduced to improve the model accuracy. The results show that the root mean square errors of the average temperature between hybrid reduced model (HRM) and HFM are 0.0399 K and 0.1989 K respectively for the cases without and with thermal management. In addition, the HRM is able to capture dynamic characteristic under perturbed input variables with a relative error of temperature within 0.4%, indicating great potential for system-level simulation. What's more, the analysis of predominant pole derived from the state matrices of different thermal managements, reveals that the expanded graphite has superior effect on the adjustment time extension compared with fins, while the fins have distinct advantages in heat transfer enhancement at the cost of volumetric hydrogen storage density.